Probability for bribe payment: multi-country approach for Latin America and the Caribbean

Abstract

Bribery, as research issue, focuses on causality and impact analysis and in many cases has been subordinated as one of the many types of corruption. Using microdata from Transparency International's Global Corruption Barometer, the objective is to determine for a set of selected countries in Latin America and the Caribbean, the dependent variables to predict the probability of paying bribes when in contact with individuals, companies or entities that provide services. Inferential statistical techniques are used, such as logistic regression applying the maximum likelihood method and iterative process until the best model is found. From an initial survey with 117 variables, 11 variables were found to best explain the probability of paying a bribe. Variables such as level of schooling and the influence of officials and politicians affect people's willingness to pay a bribe. Significant topics of this study, such as retaliation to favor voting and sexual favors to receive benefits, although incipient in their study, are worth further research. This work can serve as a basis to motivate studies with microdata that help to better understand specific typologies of corruption through models that predict behaviors of ordinary citizens.

Keywords

• américas
• corrupción
• probabilidad estadística
• regresión logística
• soborno